100G network core, 1.2 Tbps of capacity
Our 100G network core handles up to 1.2 Tbps of switching capacity. Internally, bandwidth is not the limiting factor: we keep a large reserve to grow the network on top of it.
100G network core (1.2 Tbps of capacity), scalable 10G uplink, PoP in Nozay (44), Netrix 24 Tbps+ anti-DDoS included and recent 100% NVMe hardware on passive cooling. We explain exactly what runs your servers, with no empty marketing.
Bandwidth is good. Bandwidth that holds under attack, backed by a solid operator, is better. Here's how we wired HG, with room to grow.
Our 100G network core handles up to 1.2 Tbps of switching capacity. Internally, bandwidth is not the limiting factor: we keep a large reserve to grow the network on top of it.
Our outbound link is currently 10 Gbps, plenty for today's traffic. The 100G core behind it is already ready for much more: we will scale the uplink up as load grows, without touching the internal infrastructure.
We rely on the Netrix network, a specialised French operator, for our transit and mitigation. A solid, proven partner rather than a patchwork of small links: that is what guarantees you clean, stable traffic.
SNMP / sFlow telemetry on every router, real-time throughput graphs, threshold alerting. We spot a network anomaly before you open a ticket about it.
Key figures: 100G Network core · 1.2 Tbps Switching capacity · 10 Gbps Current uplink (scalable) · 24 Tbps+ Netrix DDoS mitigation.
Our AS (AS211333) is multi-homed: through our operator SERVERD and several transit providers, we reach the major global Tier 1 networks directly. More paths means better latency, redundancy and resilience.
Indicative topology of our autonomous system AS211333 (paths change in real time according to BGP routing).
No hardware tucked away in a closet. Pro-grade power, passive cooling, controlled access: the environment your machines run in matters as much as the machines themselves.
Our point of presence is hosted in Nozay, in Loire-Atlantique, in a professional datacenter: operator-grade power, cooling and access control.
UPS and electrical backup managed by the datacenter. Your machines run on clean, stable power, not on a garage socket.
We rely on passive cooling: no loud, fragile forced ventilation, but natural, silent dissipation that keeps the hardware in its optimal range with fewer parts that fail.
Controlled site access, surveillance and traceability. Your servers are not in a garage: they are in a real, operated facility.
All traffic destined to our IPs is filtered upstream by the Netrix scrubbing network (24 Tbps+ of mitigation capacity), complemented by our in-house XDP / eBPF filtering on the most common gaming protocols.
It is included in our plans, with no option to enable and no hidden surcharge: what comes out is already clean when we feed it back to your infrastructure.
High-frequency CPUs, 100% NVMe storage, ECC memory and a virtualisation layer that doesn't sell you the same resources four times over.
Recent AMD Ryzen / EPYC and Intel chips, chosen for the single-core frequency that gaming and web workloads need. No old, underclocked Xeons.
NVMe across the entire fleet, in RAID for resilience. Access times that won't drag down your I/O even under load.
ECC RAM that corrects memory errors on the fly. Fewer silent crashes, more stability on long-running workloads.
Hypervisors built to avoid aggressive overselling. You get the share of resources advertised, not a quarter shared with 40 neighbours.
The 100G is our network core: it handles up to 1.2 Tbps of switching capacity. Our outbound uplink is currently 10 Gbps, more than enough for today's traffic, and we will scale it up as load grows since the core behind it is already ready. Client ports are 10G Best Effort depending on the plan.
One PoP, in Nozay (44), that we operate and master end to end. We prefer a concentrated, well-maintained infrastructure over a scattered network. Multi-site redundancy is part of what we will deploy as we grow; if your project requires it today, tell us and we will discuss it openly.
From mainland France, we typically measure under 10 ms to our PoP. For international (USA, Asia), geographic distance dominates. If latency is critical for your project, tell us: we will give you the real figures from your location before you commit.
Yes. All traffic destined to our IPs first goes through Netrix scrubbing (24 Tbps+), complemented by our in-house XDP / eBPF filtering on common gaming protocols. It is included, not an option to enable.
Yes. We announce our own IPv6 /44 prefix over BGP: routable IPv4 and IPv6 are provided by default on our hosting plans, at no extra cost.
Traffic volume, workload type, latency sensitivity: we look together at whether our network and hardware are built for you. Ticket or Discord, fast reply.